Folding@Home in 2010

By , the in Folding@Home Project - No comment
Many people are probably asking themselves: what's in store for the project in 2010? Well, at the end of this year, we already have some answers.

Firstly, the redesign of the psummary system just ended. The new script is ready to usher in new projects and new v5 servers. This new version of the server software comes into production after the correction of a few last minute bugs. 2010 will be the year of v5 servers, new machines and new projects; good news for the stability of the entire system.

As explained in the previous news, 2010 will also see the start of the production of a new statistics system. Currently, most of the features are disabled (such as tracking the number of folded WUs project by project, for example), but downtime during updates should rapidly disappear. Folding@Home is ready to continue its expansion.

The rewrites of many parts of the v7 Folding@Home client is expected to fill many common complaints by new and existing users. We anticipate a unified client that is easy and sleek to use and set up, without the need to carry out sometimes obscure and excessive installation procedures. Vijay Pande believes that the client should be available during 2010.

The WU assignment system should also be much improved with the rewrite of the server assignment code, which has by now become very complex, often leading to quirks and downtime. This new AS (Assignment Server) code is planned for late 2010.

We are also preparing to welcome in new computing cores: GPU3, SMP2 and Protomol. GPU3 marks the transition from proprietary GPGPU technologies (CAL/Stream for ATI and CUDA for nVidia) to a unified platform: OpenCL (through the open source OpenMM program developed by Stanford). We hope that the transition to OpenCL will improve the compatibility of computing cores with new generations of GPUs without a necessitating a complete rewrite of the code. 2010 we will also show whether or not ATI will return in the OpenCL performance race with nVidia.

Protomol, which made its appearance this season, should also take up an important place during 2010. It is one of the priority projects developed in response to a focus on performance improvements, so expect to see new projects using this core next year. Protomol might also come into play as far as GPU folding is concerned, since it is designed to interface with OpenMM.

On the multi-core CPU front, 2010 will be the year of threads. The SMP2 cores (based on Gromacs and Desmond) should be launched this year, and will excise the MPI layer and processes model. We all hope that this will simplify the use of SMP and that their clients will show more equality between Windows and Linux in terms of performance. We also assume that with these new cores will come new projects, along with the ability to engage other areas of study such as on the influenza virus (primarily the 266x and 267X projects, run under SMP).

On the occasion of its 10th anniversary, the project is slowly renewing; the older tools are gradually being revamped to support the incredible growth of the project. FAH-Addict will not fail to keep you abreast of the latest developments.